This scientific research presents the results of a survey among a population of 18 to 64 years old, of which 18 men and 32 women. A total of 50 people were interviewed. The subject of the study is family violence, the subject is the aggressive behavior of respondents in the family. Objective: to analyze the aggressive behavior of respondents in the context of the problem of family violence through a sociological survey and data processing in the PSPP program. With the help of the semantic differential, we learned the level of emotional attitude to the friendly families without conflicts, to families with conflicts and to their real family. Respondents on the proposed grounds assessed their emotional attitude to different types of families. A semantic portrait was made and primary distances were found. Mostly, 68% of respondents believe that there is a problem of violence in families, 12% do not think that there is a problem, 10% do not know, 6% and 4% do not know and do not think about it. The majority of respondents - this is 76%, quarrels in the family are rare, 10% - often, 2% - very often and 12% never.

Keywords: subjective perception, emotional states, questioning, distribution functions, comparative analysis, semantic differential, conjugacy tables

The work is devoted to one of the indicators of hidden family aggression – emotional components of family attitudes. The research is based on the pilot measurements carried out by the survey using the method of semantic differential. A total of 50 people were interviewed, including 18 men and 32 women, aged 18 to 64 years. The option with calibration was used, and respondents were asked to characterize on a semantic scale the real families and both ideal positive and negative ones. Images of ideal constructs defined in semantic space the scale and direction. In accordance with Coombs's "ideal point" approach, the level of emotional perception was estimated by us from the relative distances from the image of the real object to the images of ideal constructs. Analysis was performed for each respondent separately, and based on the results the distribution functions of emotional perception by respondents were built. They reflected the internal structure of the group for affective components. The calculations were performed in the linear approximation (for half-difference of the above-mentioned relative distances) and within the framework of the non-linear psychosemantic phenomenological model. In the last the levels of perception are given in the form of differential equations, which were solved by the Monte Carlo method with by AnyLogic multi-agent simulation package. The obtained results confirm the earlier estimates based on the averaged data. The average value of the level of emotional perception throughout the study audience is positive (0.33 on a scale from -1 to +1), which indicates the absence of aggressive behavior for the majority of respondents. A strongly negative level (from -0.5 to -1) showed only 6%, against this background, they can develop aggressive behavior. Another 6% are characterized by a slight emotional irritation in the family sphere (the level from -0.1 to -0.3), this is an intermediate group which under adverse conditions can develop into a risk group. For the rest of the study audience (88%) emotional relationships in the family are not aggressive, positive.

Keywords: semantic differential, subjective perception, latent variables, psychosemantic phenomenological model, emotional perception, multi-agent modeling, distribution function